Breast cancer is the most common malignancy among women in the Netherlands, with a cumulative risk by age 85 of one in 11. The strongest epidemiological risk factor for the disease is a positive family history. Depending on the age of diagnosis and occurrence of bilateral disease in the index case, first degree relatives may have a relative risk of up to 10 for developing breast cancer. In the US population, 6 to 19% of women with breast cancer have at least one affected relative at the time of diagnosis [1], but not all of them are expected to be true genetic cases as the high incidence of breast cancer in the general population will inevitably cause some coincidental familial clustering. In an attempt to stratify the two classes, criteria to define truly inherited breast cancer have been proposed [2]. Such cases are characterized by early age of onset (premenopausal), excess of bilaterality, and clear paternal or maternal transmission with an autosomal dominant mode of inheritance. Approximately 5% of all cases comply with these criteria, while another 13% are classified as familial clustering [3]. Since early age of onset appears to be a hallmark of hereditary breast cancer, one may suspect that among these cases the genetic component is much higher. Indeed, up to 35% of cases diagnosed under the age of 30 are expected to be genetic [4]. No such data are available for the Dutch situation, and little or none of this has been confirmed at the molecular genetic level.
Linkage analysis of early-onset breast cancer families localized BRCA1 to the long arm of chromosome 17 [5]. Further analyses of additional families revealed that women inheriting a mutant allele of BRCA1 are also at increased risk for ovarian cancer [6, 7]. Overall, approximately 45% of all families in which breast cancer is the predominant malignacy are due to BRCA1, as are over 80% of all families with both breast and ovarian cancer [6, 8]. Female mutation carriers have been estimated to have an 87% risk to develop breast cancer before the age of 70, and 63% risk to develop ovarian cancer before the age [7]. However, significant evidence for ovarian cancer risk heterogeneity was obtained, indicating the existence of at least two classed of BRCA1 mutations; one conferring a high risk to both breast and ovarian cancer, and one conferring a high risk to breast cancer, but only a moderate risk to ovarian cancer, with the former comprising approximately 26% of all BRCA1 mutations [9]. The gene frequency of BRCA1 has been estimated to be 1 in 833 women [10]. This would imply that 1.7% of all breast cancer patients diagnosed between age 20 and 70 are carrier of such a mutation.
The gene structure of BRCA1 was found to consist of 22 coding exons spanning >80 kb of genomic DNA [11], and encoding a 7.8 kb transcript [12]. An unusually large exon 11 of 3.4 kb comprises 61% of the coding domain. Over 900 mutations in BRCA1 have been published to date and compiled into an electronically accessible database [13]. Several characteristics stand out [14]. First, they are nearly ubiquitously distributed over the gene. Second, >85% of the mutations in the database lead to premature termination of protein translation. These include basepair substitutions leading to a stop codon, small insertions and deletions (of 1 to 40 basepairs) leading to a frame-shift, or splice-site mutations leading to deletions of complete exons and frame shifts. That these changes presumably inactivate gene function is supported by the finding that the great majority of breast and ovarian tumors that develop in BRCA1 mutation carriers show loss of the wildtype allele [15]. The relevance in terms of cancer predisposition of the missense mutations remains a matter of debate. Some of them appear rare polymorphic variants, as they are also observed in control samples. Others seem to affect critical residues, such as the cysteines in the amino-terminal ring finger domain [12], which are conserved in the mouse Brca1 sequence [16]. Third, a number of mutations have been found repeatedly, reducing the number of distinct mutations to about 150. Two of these, the 185delAG mutation and the 5382insC mutation, each represent approximately 11% of all mutations thus far reported [14]. Reconstruction of the haplotypes bearing some of the most common mutations has provided strong evidence that they have either a single or a few common ancestors and may have been present in the population already for several centuries [17-19]. Consequently, the incidence of specific mutations is strongly dependent on the population from which the breast cancer families were ascertained. Thus the 185delAG mutation was picked up mainly in families of Ashkenazi-Jewish origin [20]. The extent of the founder-effect was highlighted by the finding that approximately 1% of all Ashkenazi Jews (i.e. regardless of a positive breast cancer family history) are carrying this mutation [21,22], 8 times that of the incidence of all mutations together in the general population [10]. Specific mutations have also been recurrently detected in breast cancer families of Swedish, British, Italian, and Austrian origin [18,23-26].
Despite the vast number of BRCA1 gene changes detected to date, there remains a discrepancy between the proportion of BRCA1 mutations predicted by linkage studies [6,8], and the actual prevalence established by mutation analysis, among breast cancer families derived from a variety of ethnic backgrounds [27-31]. In general, this is explained in two ways: either a substantial number of mutations have been missed by the applied mutation screening methodology, or the genetic heterogeneity of hereditary breast cancer is significantly greater than hitherto expected.
Relatively little information of predictive value can be gleaned from the existing data. In one set of 35 kindreds with proven BRCA1 mutations from the United Kingdom, the ovarian cancer risk heterogeneity as predicted from linkage studies could be confirmed [25]. Mutations occurring before codon 1435 conferred a significantly higher ovarian cancer risk than those occurring after this point. While this is consistent with earlier predictions based on linkage analysis [9], the current mutation distribution is at odds with the predicted lower frequency of these alleles. In addition, the expressivity of BRCA1 displays considerable inter-family variability. For example, the 185delAG mutation was detected in families with early-onset breast cancer and ovarian cancer, or late-onset breast cancer without ovarian cancer [32]. Clearly, other factors influence the expression of the phenotype, and some of those might be genetic, others environmental. Of note, BRCA1 carriers who have a rare allele at the HRAS1 minisatellite locus were recently shown to be at a 2.8-fold increased risk for ovarian cancer relative to those carriers who had common alleles at HRAS1 [33]. However, a firm establishment of the full spectrum of BRCA1 gene changes in the population is pivotal for a more formal analysis of this matter.
An intriguing feature of BRCA1, and unexpected in the light of Knudson's two-hit inactivation theorem for tumor suppressor genes, is that somatically acquired mutations are extremely rare in ovarian tumors [34-38] and have in fact not yet been detected in 135 breast tumors [39-40]. This might indicate that inactivation of BRCA1 is not selected for during tumorigenesis of the non-inherited form of breast cancer. BRCA1 expression might be critical only during certain stages of tissue development, e.g., during puberty when the breast undergoes its final differentiation into a potential milk-producing gland [39]. However, others have argued that the mechanism of inactivation might be different from that seen in inherited cases [41].